Error using svmclassify The number of columns in TEST and training data must be equal. Error in multisvm classes = svmclassif​y(svmStruc​t,tst);

when i try to classify i got the above error in multisvm this is the code that am using

function [itrfin] = multisvm( T,C,test )
  itrind=size(test,1);
  itrfin=[];
  Cb=C;
  Tb=T;
for tempind=1:itrind
    tst=test(tempind,:);
    C=Cb;
    T=Tb;
    u=unique(C);
    N=length(u);
    c4=[];
    c3=[];
    j=1;
    k=1;
    if(N>2)
        itr=1;
        classes=0;
        cond=max(C)-min(C);
        while((classes~=1)&&(itr<=length(u))&& size(C,2)>1 && cond>0)
            c1=(C==u(itr));
            newClass=c1;
             svmStruct = svmtrain(T,newClass);            
            classes = svmclassify(svmStruct,tst);
            for i=1:size(newClass,2)
                if newClass(1,i)==0;
                    c3(k,:)=T(i,:);
                    k=k+1;
                end
            end
        T=c3;
        c3=[];
        k=1;

              for i=1:size(newClass,2)
                  if newClass(1,i)==0;
                      c4(1,j)=C(1,i);
                      j=j+1;
                  end
              end
          C=c4;
          c4=[];
          j=1;

          cond=max(C)-min(C); 
              if classes~=1
                  itr=itr+1;
              end    
          end
      end

valt=Cb==u(itr);    
val=unique(val);
itrfin(tempind,:)=val;  
end

end

Answer 1

[Walter Roberson]

tst=test(tempind,:)

has multiple columns

        itr=1;
        [...]
            c1=(C==u(itr));
            newClass=c1;

itr is a scalar, so u(itr) is a scalar, so c1 is a scalar so newClass is a scalar

             svmStruct = svmtrain(T,newClass);

You are training on T as if everything is in the same class. If you get different results for newClass = false versus newClass = true then out would be due to round-off error in the calculations. svm does not care what the one class is labeled with; it only cares that you are telling it that all of the data is in the same class.

"The number of columns in TEST and training data must be equal."

size(T,2) is not the same as size(test,2) . You did not happen to show us the sizes.

Be careful with whether your rows represent different attributes within one observation, or whether your rows represent the same attribute for each different observation. Different classification routines have different rules about whether they expect row-oriented or column-oriented values.

[Melaku Eneayehu]

This is the code that am trying to classify i used GLCM feature extraction my train_feat is 157*8 and label is 157*1 vector matrix, what happen even when i try the training image for testing always result return 1 and i gt the same result benign or the result in class 1 this is the code i used for testing, why result always return 1? the classification code is the above(multisvm)

function browseButtonPushed(app, event)
               %             global im im2
            [path,user_cance]=imgetfile();
            app.UIFigure.Visible = 'off';
            app.UIFigure.Visible = 'on';

              if user_cance
      msgbox(sprintf('error'),'error','error');
      return;
              end
              im=imread(path);
              im2=im;
               imshow(im2,'Parent',app.UIAxes)
              [rows, columns, numberOfColorChannels] = size(im2);
              if numberOfColorChannels >1
              %   % It's not really gray scale like we expected - it's color.
              %   % Convert it to gray scale by taking only the green channel.
              %   grayImage = grayImage(:, :, 2); % Take green channel.

              uiwait(msgbox('Image is Not GrayScale Conversion is needed','Attention','warn'));
              else
                 uiwait(msgbox('Image is  GrayScale Conversion is not needed','Attention','modal')); 
              end
  end 

   function classifyButtonPushed(app, event)
  img=getimage(app.UIAxes);
              glcms = graycomatrix(img);
              % Derive Statistics from GLCM
              stats = graycoprops(glcms,'Contrast Correlation Energy Homogeneity');
              Contrast = stats.Contrast;
              Correlation = stats.Correlation;
              Energy = stats.Energy;
              Homogeneity = stats.Homogeneity; 
              Mean = mean2(img);
              Standard_Deviation = std2(img);
              Entropy = entropy(img);
              RMS = mean2(rms(img));
              %Skewness = skewness(img)
              feat_disease = [Contrast,Correlation,Energy,Homogeneity, Mean, Standard_Deviation, Entropy, RMS];
               test=  feat_disease;
              % Load All The Features
              load('Training_Data.mat');
              % Put the test features into variable 'test'

              result = multisvm(Train_Feat,Train_Label,test);
              %disp(result);

              % Visualize Results
              if result == 1
               helpdlg(' Begnin ');
      disp(' Begnin ');
              elseif result == 2

              helpdlg('Normal ');
      disp('Normal');
              elseif result == 0

             helpdlg(' Malignat ');
      disp(' Malignat ');

              end
  end

[Melaku Eneayehu]

and here is the code am using to prepare Train_Feat feature matrix but i prepared the Train_Lable manually through excel because i know the label of each image first 54 class 0 second 49 for class 1 last 54 image class 2 totally 157 rows (always result is return class 1)and converted to .mat file.

clc
close all
clear all

setDir  = fullfile('./Training');
imgSets = imageSet(setDir, 'recursive');

h = waitbar(0,'Please wait...');
ki=1;
for i=1:length(imgSets)
for j=1:imgSets(i).Count
img=imresize(imread(cell2mat(imgSets(i).ImageLocation(j))),[512 512]);

[a b c]=size(img);
if c==3
    img=rgb2gray(img);
end
feat=graycoprops(img,{'contrast','Correlation','Energy','homogeneity'});
Mean = mean2(img);
Standard_Deviation = std2(img);
            Entropy = entropy(img);
            RMS = mean2(rms(img));
feature=[feat.Contrast feat.Correlation feat.Energy feat.Homogeneity Mean Standard_Deviation Entropy RMS];
dataset(ki, :) = [feature ki];
ki=ki+1
waitbar(ki/150)
end
end
close(h);
save('Training_Feat.mat','dataset')

Answer 2

[thejas A V]

use equal number of dataset and corresponding label pls refer to attachment

[Walter Roberson]

Is that a question or a response to the question ?